Improving x‐ray fluorescence signal for benchtop polychromatic cone‐beam x‐ray fluorescence computed tomography by incident x‐ray spectrum optimization: A Monte Carlo study

Abstract

Purpose: To develop an accurate and comprehensive Monte Carlo (MC) model of an experimental benchtop polychromatic cone-beam x-ray fluorescence computed tomography (XFCT) setup and apply this MC model to optimize incident x-ray spectrum for improving production/detection of x-ray fluorescence photons from gold nanoparticles (GNPs). Methods: A detailed MC model, based on an experimental XFCT system, was created using the Monte Carlo N-Particle (MCNP) transport code. The model was validated by comparing MC results including x-ray fluorescence (XRF) and scatter photon spectra with measured data obtained under identical conditions using 105 kVp cone-beam x-rays filtered by either 1 mm of lead (Pb) or 0.9 mm of tin (Sn). After validation, the model was used to investigate the effects of additional filtration of the incident beam with Pb and Sn. Supplementary incident x-ray spectra, representing heavier filtration (Pb: 2 and 3 mm; Sn: 1, 2, and 3 mm) were computationally generated and used with the model to obtain XRF/scatter spectra. Quasimonochromatic incident x-ray spectra (81, 85, 90, 95, and 100 keV with 10 keV full width at half maximum) were also investigated to determine the ideal energy for distinguishing gold XRF signal from the scatter background. Fluorescence signal-to-dose ratio (FSDR) and fluorescence-normalized scan time (FNST) were used as metrics to assess results. Results: Calculated XRF/scatter spectra for 1-mm Pb and 0.9-mm Sn filters matched (r ≥ 0.996) experimental measurements. Calculated spectra representing additional filtration for both filter materials showed that the spectral hardening improved the FSDR at the expense of requiring a much longer FNST. In general, using Sn instead of Pb, at a given filter thickness, allowed an increase of up to 20% in FSDR, more prominent gold XRF peaks, and up to an order of magnitude decrease in FNST. Simulations using quasimonochromatic spectra suggested that increasing source x-ray energy, in the investigated range of 81-100 keV, increased the FSDR up to a factor of 20, compared to 1 mm Pb, and further facilitated separation of gold XRF peaks from the scatter background. Conclusions: A detailed MC model of an experimental benchtop XFCT system has been developed and validated. In exemplary calculations to illustrate the usefulness of this model, it was shown that potential use of quasimonochromatic spectra or judicious choice of filter material/thickness to tailor the spectrum of a polychromatic x-ray source can significantly improve the performance of benchtop XFCT, while considering trade-offs between FSDR and FNST. As demonstrated, the current MC model is a reliable and powerful computational tool that can greatly expedite the further development of a benchtop XFCT system for routine preclinical molecular imaging with GNPs and other metal probes. C

Quantitative imaging of gold nanoparticle distribution in a tumor-bearing mouse using benchtop x-ray fluorescence computed tomography

“…Meanwhile, we have developed a cone beam implementation of XFCT 23 24 that, while technically more complex 25 , offers distinct advantages over the pencil beam approach, such as parallel XRF signal acquisition, and is also absolutely crucial for making XFCT suitable for in vivo imaging under the practical constraints of x-ray dose and scan time. In a more recent study, we investigated optimization of the incident polychromatic x-ray spectrum (e.g., filtration and quasi-monochromatization) 26 as it pertains to the interplay among XRF signal production/detection, dose, and overall scanning time, in order to expedite the development of a benchtop XFCT system capable of in vivo imaging. The basic concepts and an overall workflow for benchtop XFCT with GNPs are illustrated in Fig.…”

Optimal X-ray filters for EDXRF analysis

“…Using a filter for the primary X-ray beam in order to optimize the signal-to-background ratio for EDXRF measurements is reported in several publications [5,6,7,8,9]. Gilmore et al [5] demonstrated an improved detectability of As in a hydrocarbon target using a 40 µm thick Zn filter and a * daniel.maier@cea.fr target (Z T , d T ) X-ray beam Figure 1: Principal geometry of an EDXRF setup using a beam filter.…”

“…Within this range, they demonstrated that the detection limit decreases as the filter thickness increases. A Monte Carlo study by Manohar et al [8] compared a 1, 2, and 3 mm thick Sn and a 2 and 3 mm thick Pb filtration for the detection of Au in a PMMA target using U 0 = 105 kV. They concluded that the Sn filter works better than the Pb filter and that the thickness must be chosen as a trade-off between signal-to-dose value and duration of acquisition.…”

Pinhole X-ray fluorescence imaging of gadolinium and gold nanoparticles using polychromatic X-rays: a Monte Carlo study

“…48 The MCNP code has been widely used for the simulation studies on cadmium-zinc-telluride (CZT), CdTe, and high-purity germanium (HPGe) semiconductor detectors for medical imaging. [49][50][51] The pinhole XRF imaging system has an imaging mechanism very similar to SPECT. 38 The model was benchmarked against the design and the dimension of the pinhole, as well as the type and the pixel size of the detector commonly used in preclinical pinhole SPECT.…”